India's Enterprise AI Growth Triggers Governance Focus on Agent Control

A new SAP and Oxford Economics study of 2,600 business leaders, including 200 from India, shows Indian enterprises are moving beyond AI pilots toward wider operational use. Companies plan to invest an average of US$25.9 million in AI, with spend expected to increase by 45 % over the next two years. AI currently supports about one‑third of business tasks and is projected to handle more than half of them by 2028. Around two‑thirds of Indian firms are already piloting agentic AI, and 85 % see it as a major transformation driver, though ROI expectations rise from 22 % to 39 % in the same period. While 63 % consider their data ready, respondents cite incomplete data (76 %) and data‑quality concerns (67 %) as major barriers, and only a small share feel fully prepared in skills (11 %) or governance (14 %).

At the same time, experts warn that the next challenge for enterprise AI is not more agents but control. Autonomous AI agents now access documents, tools, and workflows, creating new risk profiles. Effective governance requires visibility into all deployed agents, clear identity and permission models, runtime enforcement, and auditable action logs. Existing mechanisms such as OAuth, designed for human‑approved applications, are insufficient on their own. Companies must inventory agents, define ownership, scope access, and ensure that agents can be revoked instantly without disrupting operations.